could also be more playful, like having little knobs to turn to tune different parameters, or totally rearrange things when you click a button. more video game, less assembly line

interesting topic ... im getting something like on one hand it is just a new kind of uncertainty ... and also different types of uncertainties have different qualities

scamming as an epistemic comparison brings things up for me, because even as we learn to avoid scams, we also learn to be more guarded in places where scams occur (e.g. train stations)

half-related to recent thread by @aparrish about closed epistemic loops friend.camp/@aparrish/11...

like yes humans are used to navigating unreliability, but llm unreliability has a specific flavor that i find can sometimes be particularly misleading in a way that shitty internet articles and stack overflow posts never were for me personally. like no one on SO ever fabricates a very specific api

I largely agree with the quoted tweet and your addition, but I have personally found the way that llms do still by their nature sometimes confidently fabricate very specific and incorrect things, to be a kind of sneaky epistemic tarpit and can be really misleading for topics i dont already know well

this is clever and cute and also fair hahah